Elon Musk has made a strikingly honest admission, saying that his own Grok 4.8 AI model is “not as good” as Anthropic’s Claude Opus 5.5.
The remark landed after a rough week for Grok Bot. Users said their allowances ran out unusually fast. On Nov. 1, Musk wrote that “the cost of serving Grok Bot was unsustainably high relative to revenue.” On Sunday he said, “Grok Bot fixes for improved token efficiency and smarter behavior are underway.” Matt Palmer, who works on the product at SpaceXSI, quoted that post and asked users for feedback. The official @bot account reset weekly limits for all users on Monday, and in a Tuesday video Palmer said the team had been “working all weekend on it,” as WebProNews reported.
Musk was answering a question about how Grok 4.8, which he said would get a soft rollout to Bot users later in the week, would compare with Opus 5.5. Days after calling Grok Bot’s serving costs unsustainable, he still put a rival model ahead of his own. The praise, and the decision to reset limits and keep fixing the product rather than starve it of usage, points to SpaceXSI putting product quality ahead of cost savings.
X’s Grok is a popular alternative to OpenAI’s ChatGPT and Anthropic’s Claude. Musk and company have tried to position Grok as a less politically correct, more brutally honest chatbot. In addition, Grok has access to all of X, meaning it is outstanding at having the pulse of social discourse.
Despite its strengths, and the strides it has made in recent versions, Musk says there’s no comparison between Opus 5.5 and Grok.
While it’s entirely possible Grok 5 will surpass Opus 5.5, it’s also a safe bet that Anthropic will have released an improved version of their model in two or three months’ time.